I'm working on a story universe that takes place mostly/entirely within a multi-layered Dyson sphere. The story is not hard sci-fi, but I am trying to ensure the physics or sci-fi elements "feel" like they make sense, even if they wouldn't pass a physicist's analysis. The layout of the sphere is:

  • A white dwarf star/artificial star core, a little bigger than Earth. The shell around the star is made entirely of handwavium sci-fi solar panels to collect all the energy needed for the rest of the construct.

  • Each successive layer outward from the core is a habitable zone where life forms can safely exist. I'm not certain exactly how many layers there are, but I do know that the layers are "thick" enough to provide atmospheres of at least equivalent to the Earth's. In other words, the "ceiling" of each layered habitat is at least 10,000 km above the "ground".

  • Each layer will have its own ecology and day/night cycle. My question is twofold...

    1. Could the "ceiling" of each Dyson shell layer have its own sun/sunlamp that cast enough light to illuminate an entire hemisphere (and thus rotate around on a rail system of some kind, sort of like a sun-moon light source)? I am unsure since the sunlamp/light source would be much closer to the hab-layer's inhabitants than our own sun, so I don't know if the distance would be far enough to light everything.

    2. If 1. is not possible, how best would the sphere's creators go about lighting everything in each hab-layer? Would there be multiple suns/sunlamps in a given hemisphere that rotate around together to light the sphere in a sequence? A light band or line instead of a sun? Something else?

Thanks for any insights you can provide!

  • $\begingroup$ Please clarify your specific problem or provide additional details to highlight exactly what you need. As it's currently written, it's hard to tell exactly what you're asking. $\endgroup$
    – Community Bot
    Commented Jan 16, 2023 at 17:45
  • 1
    $\begingroup$ /the layers are "thick" enough to provide atmospheres of at least equivalent to the Earth's. In other words, the "ceiling" of each layered habitat is at least 10,000 km above the "ground"./ you could have such an atmosphere with less space and more pressurizaton. Like on the space station. $\endgroup$
    – Willk
    Commented Jan 16, 2023 at 17:48
  • $\begingroup$ I could, but part of the story's gimmick is that each layer is a world unto itself, so there's enough room for characters to explore, aircraft, super high mountains, etc. The layers "need" to be quite thick for the kind of world I'm thinking of. $\endgroup$
    – AWryBread
    Commented Jan 16, 2023 at 17:55
  • $\begingroup$ @AWryBread 10000km is a long way though. That's a lot of boring empty vacuum where you can't do a whole lot of interesting stuff. $\endgroup$ Commented Jan 16, 2023 at 18:13
  • $\begingroup$ That's a good point. May not really be necessary to be 10,000 km high. That said, if the "ceilings" are lower than that my issue about whether or not an artificial "sun" is high enough to illuminate a hemisphere is even worse! $\endgroup$
    – AWryBread
    Commented Jan 16, 2023 at 18:21

4 Answers 4


/day/night cycle/

Many individual fixed lights, each getting brighter and dimmer over a day.

Moving a light is a clumsy way to have it go out. They have plugs and stuff and you are liable to trip over the dog.

Instead, you can have the light go out by turning it off. This could be done by a "Clapper" like system with one of your worlds inhabitants elected as official clapper. Or with a big switch.

Or best: have the lights on a timer and when it is the right time they slowly turn on. Your lights are positioned on a circumferential band around the ceiling of the world. As the day progresses each light slowly turns on, increases to maximal brightness then slowly dims down and turns off. The result is a "sun" comprised of several adjacent lights which appears to move across the sky over the course of a day. Really the lights are just coming on and turning off.

Sometimes one gets stuck on. Or flickers, which makes the dogs bark because it looks weird. That means the bulb needs to be changed.

It occurs to me that given the lights are closer to the ground n this world than the sun is to us, it might be necessary to have several sun bands each illuminating a different latitude. Most places will get light from more than one sun at once, with the main one or 2 lights more or less overhead and adjacent suns closer to the horizon. They all change together.

  • $\begingroup$ This is a very good idea - I had thought maybe having multiple "suns" might be the ideal solution, and having several lights illuminate the landscape could make for some interesting visuals and world elements. $\endgroup$
    – AWryBread
    Commented Jan 16, 2023 at 18:20

I would reference Iain Banks's book "Matter", which takes place on a planet made of numerous concentric shells, but planet-shells instead of Dyson shells. The inner shells are lit and heated by "roller stars" that are fusion reactors that roll across the ceiling.

It sounds counter-productive to put smaller fusion reactors inside of a Dyson shell. The entire point of a Dyson shell is to maximize use of the parent star. If you have a layer that can't see the parent star, then you might as well build space stations.

I also want to suggest that, while your setting sounds interesting, I think you severely under-imagine how much surface area a Dyson sphere has, although building one around a white dwarf is a cool innovation. I saw a calculation that the goldilocks zone of a white dwarf would be somewhere in the 1-2 million mile range. Even at the lower bound, the surface area is equivalent to ten million Earths.

  • $\begingroup$ Thanks, I've seen references to Matter as well. I'll have to dig a bit deeper there. $\endgroup$
    – AWryBread
    Commented Jan 16, 2023 at 19:48

Dyson Onions huh? That's fun on multiple levels, for some occasionally surprising reasons. For instance, did you know that gravity inside a Dyson sphere is towards the mass at the center? That means that in any pair of shells the 'ground' will be the outside of the inner shell... which actually affects the answer quite a bit when you think about it. The gravity isn't huge - around 1/1650th of Earth gravity at 1AU from our own star if I've done the math right - but it's definitely in the direction of the star.

Anyway, the other thing that matters quite a lot is the geometric radius (ignoring space warping around the star's mass, because it's inconvenient) of the ground sphere. This is important for calculating a tangent between an arbitrary point on the sky sphere, which we can use to approximate the distance on the ground sphere that any particular light will be visible from. It will also give us the distance a 'sun' would have to travel from one horizon to another, which in turn will tell us how fast that 'sun' will have to travel and, eventually, how many of those 'suns' you're going to need.

What we need is a way to easily calculate a chord across the equator of the sphere:

Chord of a Circle

That's a nice little calculator because it will fill things in as long as it has a minimum of information, so let's try some numbers. Let's start with an inner radius of 0.5 AU, or around 75 million km. Putting 75,010,000 (inner radius of outer shell) into the radius and 10,000 into the chord height and... ooh, look, we have lots of pretty numbers now.

Chord length is 2,449,571 km and the arc covers 1.871 degrees. We can double the angle to account for day + night, then divide 360 by that to get the number of lights we need at the equator to provide a symmetric day/night cycle. That boils down to a little over 96 light sources. And since they have to travel the arc length in about 12 hours, each light source is moving at around 200,000 km/hr. Oh, that's pretty darned fast.

How about 0.1 AU? Let's see... 15,010,000 radius, 10,000 height still... carry the one... 43 lights travelling at ~91,300 km/hr. Well, that's slightly better... at least that's less than 10 times as fast as the ISS.

But of course that's just a narrow strip around the equator of the shell dealt with. You're going to need multiple rings of lights, and they have to not overlap so it'll have to be a latitude arrangement, and the speeds all change, and tesselation is weird, and...

You get the point, hopefully. Nope, this is A Bad Idea.

So I think (or at least hope) that we've put the mobile sun idea to rest. Even ignoring pure physics it's a bit hard to argue with the plain geometry of the thing.

So instead of mobile lights, how about a tesselated array of light panels covering the entire ceiling? The lights wouldn't have to rush around, just turn on and off at the right times. You could have half of the inner surface lit at any one time, with some of the panels brighter to simulate the presence of a sun-like light source. And this way you can vary the light durations over the 'year' to provide seasonal variances and stuff without having to constantly change the speed and direction of your lights.

I mean, sure, we're talking about a few tens of quadrillions of square kilometers of surface area at 0.5 AU (~7e+16, or ~2.8e+15 at 0.1 AU), but that's what you get for deciding to engineer a Dyson Onion.

  • $\begingroup$ Heh, yeah, it seems I've greatly underestimated the difficulties inherent in a semi-realistic Dyson Onion. $\endgroup$
    – AWryBread
    Commented Jan 17, 2023 at 16:13

No... Well, kinda....

Let's set aside the specifics of your sphere for a moment and use the following assumptions:

  • Dyson sphere around Earth's sun.
  • Dyson sphere's average radius is 1 AU.
  • Dyson sphere has four habitable layers.

Can we illuminate one hemisphere of each layer for half of every 24-hour period such that the layer experiences the same energy striking the ground (for plant growing purposes) as would normally occur on Earth?

Answer: No.

There isn't enough light/energy to work with. You're consuming all of it and then some. If you try to illuminate two layers at the luminosity experienced by Earth (the whole layer, not a hemisphere, but only 50% of the time for each layer), then you'd need the entire sun to do it. That's the problem. The surface area of one of those layers is breathtakingly enormous — but the only way to get the per-square-meter energy experience you find on the day side of Earth is to use 100% of the sun to do it.

Which means you don't have power to do anything else. (We'll get to this later.)

Your conditions make everything worse

Now let's bring your specifics into play.

  • You don't have a lovely main sequence yellow dwarf star like our Sun. You have an aging white dwarf star that emits less energy.

  • You're trying to mimic the atmosphere from the ground to space (10,000 km) rather than the biosphere (20 km). I'm not at all sure why you're doing that, but it increases the surface area something awful, thereby reducing the density of redistributed energy from the star.

  • You haven't defined the number of layers, but I'm assuming it's more than two.

The solution is that it's all about energy density

The smaller the diameter of your sphere and its layers, the greater the energy density you have to work with. If you move Earth closer to the sun, it eventually burns up because the energy density increases. This means the distance between the white dwarf star and your solar panel layer must be pretty small (in terms of AU...) to gather enough energy that the next layer (and this is the important part) can be sufficiently illuminated.

A simple way of thinking about this (it's not quite correct, it's just a simplification) is that if you want to illuminate ten layers to Earth-like standards, then you need those layers around Sol to begin at 0.2 AU. If you want 20 layers, then the average distance must be 0.1 AU. Etc.

And you don't want them 10,000 km high. There's no point for that, is there? You're not trying to let your inhabitants forget they're on an artificial sphere and try to put satellites into orbit within their layers, are you? Engineering is about efficiency. You don't do anything wasteful unless there's a whomping good reason for it. 20 km. that's what you want. Plenty of space for birds and airplanes.

Whatever you do, give up at least one layer worth of energy just to operate the sphere

For the sake of believability, you can't chew up all the star's energy for illumination. Some trivial part of it will be used by the sphere's inhabitants for disco lighting, loud music and cooking BBQ. But more to the point, you'll be processing water (a LOT of it), processing air (a LOT of it), keeping the sphere in alignment with the star, dealing with meteor defense, and a lot of other fairly energy intensive things. For the sake of believability, you want at least one layer's worth of energy to be available for those other tasks.


So, "realistically" what you're trying to do is limited. In your case, the number of layers might only be four with the internal radius of the sphere sized for five so you have that extra energy. I haven't done the math, but this might mean you're sphere is within 0.05 AU of the star (Mercury's orbit is approximately 0.4 AU...). You'll need to compare the energy out put of a white dwarf compared to good old Sol to figure out the specifics.

One last thing...

Many people who play with the idea of dyson spheres think fairly simplistically in terms of solar panels. They wouldn't be like solar panels you find here on Earth. We have those useful Van Allen radiation belts and a thriving magnetosphere that keep a lot of the really nasty photons off the planet. You can handwave that, but it's useful to calculate the entire energy output of a star and not just the energy that impacts the Earth. You have all that to work with... if your solar panels have the juice to do it. Hard radiation isn't your friend, but can be very useful.

  • 1
    $\begingroup$ "Hard radiation isn't your friend" is a bit of an understatement. Personally I'd be going with a multi-stage solution: meta-material photon collection, phase-change heat engines (steam turbines basically), particle traps with more heat exchange generation, thermocouple effect and big fat heat pipes to drain the excess to massive heat sinks on the outside. Maybe siphon some of the photons through optical channels to power the lights too. $\endgroup$
    – Corey
    Commented Jan 17, 2023 at 6:11
  • $\begingroup$ Very informative comment, especially about the limits of the sphere's design. I'll have to consider all this. $\endgroup$
    – AWryBread
    Commented Jan 17, 2023 at 16:12

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .